Stirling's formula

Stirling's formula
Name	Stirling's formula
Field	Mathematics
Introduced	18th century
Mathematician	James Stirling

Stirling's formula

Stirling's formula gives an asymptotic approximation for the factorial function and the Gamma function, relating large integers to exponential, power, and square-root terms. It connects foundational results in Leonhard Euler's work on the Gamma function with later developments by James Stirling and others, and plays a central role in analysis, probability, combinatorics, and statistical mechanics. The formula appears across literature tied to figures such as Pierre-Simon Laplace, Adrien-Marie Legendre, Carl Friedrich Gauss, and Pafnuty Chebyshev.

Statement

The classical asymptotic statement asserts that for the factorial n! and the Gamma function Γ(z) as z → ∞ along the positive real axis, one has n! ~ n^n e^{-n} sqrt{2π n}, and more generally Γ(z + 1) ~ z^z e^{-z} sqrt{2π z}. This form is associated historically with James Stirling and refined by Abraham de Moivre; it features the constant sqrt{2π}, linked to Bayes-related integral approximations used by Pierre-Simon Laplace and the Central limit theorem in the work of Carl Friedrich Gauss and Andrey Kolmogorov. Alternate presentations replace sqrt{2π} by an asymptotic series involving Bernoulli numbers first systematized by Jakob Bernoulli and later organized by Adrien-Marie Legendre.

History and derivation

Early approximations to factorials trace to Abraham de Moivre's 1730s work on probability and the normal approximation to binomial distributions; de Moivre's asymptotic expansion anticipated the formula. James Stirling published a closed form in the 1730s in correspondence with John Wallis and refined by exchange with Leonhard Euler. Subsequent treatment by Pierre-Simon Laplace connected the formula to his method of steepest descent and integrals studied in École Polytechnique-era analysis. The derivation tradition continued through contributors such as Adrien-Marie Legendre, Carl Gustav Jakob Jacobi, Niels Henrik Abel, and Srinivasa Ramanujan, each linking combinatorial enumeration, analytic continuation of the Gamma function, and asymptotic techniques that appeared also in the works of Henri Poincaré and Sophie Germain.

Proofs and refinements

Multiple rigorous proofs arise from diverse analytic tools: Laplace's method, the Euler–Maclaurin summation formula, the saddle-point method, and complex analysis via contour integration. Notable refinements include the full asymptotic expansion using Bernoulli numbers and the Stirling series, developed by Jakob Bernoulli, expanded by Adrien-Marie Legendre, and employed in modern analytic number theory by G. H. Hardy and John Edensor Littlewood. Alternate proofs invoke inequalities due to Pafnuty Chebyshev and monotonicity arguments used by Émile Borel and Harold Davenport. Modern expositions apply methods from Ernst Kummer and Bernhard Riemann's techniques for the zeta function, and links to Paul Erdős's combinatorial estimates and Atle Selberg's analytic methods appear in contemporary literature.

Applications

Stirling-type approximations are indispensable in asymptotic enumeration in combinatorics (e.g., Paul Erdős, Graham–Knuth–Patashnik contexts), statistical physics (e.g., Ludwig Boltzmann, Josiah Willard Gibbs), information theory (e.g., Claude Shannon), and Bayesian statistics (e.g., Thomas Bayes, Pierre-Simon Laplace). They underpin approximations in random matrix theory linked to Freeman Dyson and Tracy–Widom distribution studies, in approximate counting problems treated by Donald Knuth and Persi Diaconis, and in asymptotic estimates in analytic number theory connected to Godfrey Harold Hardy and Srinivasa Ramanujan. Stirling-type formulas are used in algorithm analysis in computer science influenced by Alan Turing and John von Neumann, and in thermodynamics where James Clerk Maxwell and Ludwig Boltzmann used factorial approximations for microstate counting.

Generalizations and extensions

Extensions generalize to complex arguments via the Gamma function, to Barnes G-function studied by Ernest William Barnes, and to multivariate analogues such as the asymptotics of the multivariate Gamma function used in work by Harold Jeffreys and David Slepian. Refinements include the Stirling series with Bernoulli numbers, uniform asymptotic expansions for large complex arguments linked to Frank W. J. Olver and NIST, and q-analogues connected to Leonard Euler-type q-series and work by George Gasper and Mizan Rahman. Matrix and operator extensions appear in random matrix theory by Tracy–Widom-related research and in representation-theoretic contexts investigated by Harish-Chandra.

Error bounds and numerical accuracy

Rigorous error bounds derive from the Euler–Maclaurin formula and saddle-point estimates, with classical explicit inequalities due to Pafnuty Chebyshev and later sharpenings by I. N. Bronshtein and N. M. Temme. Modern numerical analysis treatments cite controlled remainders in the asymptotic series using results associated with Frank W. J. Olver and error constants traceable to Adrien-Marie Legendre's original refinements. High-precision computational routines in libraries influenced by Donald Knuth and implemented in projects akin to GNU-based numerical packages use Stirling-based approximations with corrective rational approximants to ensure stability for moderate n and complex z, matching error analyses appearing in the work of John von Neumann and Alan Turing on numerical methods.

Category:Asymptotic analysisCategory:Gamma function